Skip to content

Parameters vs. Context: TRACE Fine-Tuning for Robust Retrieval-Augmented Generation

Sep 2026 · arXiv.org · Vol abs/2609.30337 · 0 citations · 35 references
Computer Science

TL;DR

Experiments across multiple knowledge-conflict scenarios and datasets show that TRACE improves robustness against misleading retrieved knowledge and reduces incomplete answers, and jointly enhance knowledge-source selection, conflict robustness, and answer quality in RAG models.

Abstract

Retrieval-Augmented Generation (RAG) mitigates knowledge obsolescence and factual hallucination in large language models by introducing external context. However, when retrieved knowledge conflicts with the model's internal parametric knowledge, the model may either blindly follow misleading context or incorrectly rely on parametric knowledge, leading to unreliable responses. To address this issue, this paper proposes TRACE (Debate-TRace and Answer-Completeness rEgularized fine-tuning), a robust fine-tuning framework for RAG under knowledge conflicts. First, we propose a fine-tuning method that leverages multi-agent debate traces to extract correct candidates, incorrect candidates, and answer-shift patterns, providing fine-grained supervision for reliable knowledge-source selection. In addition, we design an answer completeness regularization mechanism to alleviate empty, overly short, and prematurely terminated responses via answer-tail token reinforcement and premature termination suppression. The fine-tuning objective combines correct-answer supervision, incorrect-candidate suppression, answer-tail token reinforcement, and premature termination suppression, enabling the model to use reliable external context, resist misleading or irrelevant retrieved content, and fall back to parametric knowledge when retrieved evidence is unreliable. Experiments across multiple knowledge-conflict scenarios and datasets show that TRACE improves robustness against misleading retrieved knowledge and reduces incomplete answers. These results demonstrate that multi-agent debate traces and answer completeness regularization jointly enhance knowledge-source selection, conflict robustness, and answer quality in RAG models. Our code is available at https://github.com/PHD-lanyu/TRACE.

View source

Similar papers

Preprint Aug 2026

Finding the Right Evidence: Factor-Guided Coarse-to-Fine Reasoning for Long Videos

Consistent gains over DVD on LVBench, Video-MME, EgoSchema, and LongVideoBench suggest that option-aware evidence acquisition transfers beyond MMR-V, and proposes PACE (Progressive Acquisition of Critical Evidence), a factor-guided framework for long-video evidence acquisition.

Bai-Xuan Xu, Yinyui Xu, Tianshi ZHENG et al. · 2 citations
Review Open access Aug 2026

Large Language Models Hallucinate and How Retrieval- Augmented Generation Mitigates It

It is concluded that RAG should be viewed as a grounding and evidence-access mechanism rather than a guarantee of hallucination-free generation, and applications of RAG are outlined.

Shyalaja L. N., Shantinath Patil, P. R. et al. · 0 citations
Open access Aug 2026

Parameter-Efficient Contextual Calibration for Hallucination Mitigation in Domain-Specific Large Language Model Retrieval-Augmented Generation

CAL-RAG (Context-Aware Low-Rank Calibration for RAG), a parameter-efficient fine-tuning and decoding calibration framework designed to enforce strict contextual faithfulness without compromising generative fluency, is proposed.

Sophia N. Tawar, Liam K. Peing, Amani Bellow · 0 citations
Preprint Aug 2026

Why RAGs Hallucinate: Penalty-Aware Evaluation of Retrieval-Augmented Generation Systems with Knowledge-Gap Canaries

A penalty-aware evaluation framework for deployed RAG products is presented, combining asymmetric scoring, knowledge-gap canaries, questions whose answers are verifiably absent from the knowledge base, so that any answer constitutes ungrounded generation from parametric memory, and a failure-attribution pipeline that s...

Alden Do Rosario, Hussein Younes, F. Pires · 1 citation
#natural language process... Preprint Aug 2026

Learning to Reason and Use Tools through Unsupervised Fine-Tuning in Task-Oriented Dialog Systems

An unsupervised fine-tuning pipeline that harvests reasoning trajectories via in-context learning inference via in-context learning inference is proposed, enabling Large Language Models (LLMs) to access external knowledge and produce factual responses.

Mark A. Ferro, Oier López de Lacalle · 0 citations

Related blog posts

Microsoft Research Blog Nov 16, 2026

TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient

AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.